Real-Time Estimation of Seaway Orientation and Sea State Based on the Covariances of Submerged Body Interactions

The non-stationary ocean waves have been a subject of various engineering disciplines. A combination of different wave profiling measurements and spectral techniques has been an approach to identify the ocean wave properties. In this study, the fully appended rigid cylindrical body simulates the submerged underwater vehicle maneuvering under the seaway.. The hydrodynamic interactions between the body and wave are observed to characterize the wave effects in terms of forces and moments acting on the body. The variances and covariances of time-series measurements are examined to estimate the wave intensity and orientation in real-time. We verified that the polarities of different covariances are not only the key characteristics among different wave orientations but also well immune to the body's speed, depth as well as the strength of surface swell. The combined polarities of different covariances establish the sequential dichotomization scheme to detect the wave orientation. The wavelet decomposition of the covariances improves their polarizations by extracting only the primary wave components. Leveraging the detection of wave orientation by the wavelet coefficients of covariances, we designed the different structure Kalman filters and applied them for the different wave orientations to estimate the wave intensity in real-time. We learned that the wave intensity is relatively well measured by the variances of interacting body motions. The specific group of variances contributes the measurement vector, which is updated by the sliding-window in real-time. The uncertainties of the models are treated by adjusting the process and measurement noise variances. The simulations of the specific wave orientation modeled Kalman filters against the similar wave orientation data showed promising real-time performances. Taking away of these results, the first practical design was finalized in the sequential scheme such that estimation of sea state level becomes conditional after detecting seaway orientation in advance. The complete three dimensional variance measurements space was also developed to estimate orientation and intensity, simultaneously against the first sequential scheme utilizing two state model Kalman filter and non-parametric nearest neighbor algorithm. The study compares the estimation performances of the sequential and simultaneous designs. The study investigated the trade-off factors of computation costs, solution accuracy and convergence in real-time implementation of both designs. The sequential approach was proven to be better choice in this problem. Relaxing uncertainty from the binary decision tree to estimate the orientation classes was achieved from the investigation of Kalman innovation process especially among similar neighboring orientations to find the actual seaway heading values and their relationship to the intensity of corresponding sea state. The study offers new practical estimation approach based on the submerged body interaction frameworks focusing on the safety of underwater vehicle operation.